PDF US Report and Proposal Template

November 19, 2014

American Financial Services Association

Fair Lending:

Implications for the Indirect Auto Finance Market

Prepared by: Arthur P. Baines and Dr. Marsha J. Courchane

For: American Financial Services Association

Charles River Associates 1201 F Street, NW Suite 700 Washington DC 20004

Date: November 19, 2014

The conclusions set forth herein are based on independent research and publicly available material. The views expressed herein are the views and opinions of the authors and do not reflect or represent the views of Charles River Associates or any of the organizations with which the authors are affiliated. Any opinion expressed herein shall not amount to any form of guarantee that the authors or Charles River Associates has determined or predicted future events or circumstances and no such reliance may be inferred or implied. The authors and Charles River Associates accept no duty of care or liability of any kind whatsoever to any party, and no responsibility for damages, if any, suffered by any party as a result of decisions made, or not made, or actions taken, or not taken, based on this paper. Detailed information about Charles River Associates, a registered trade name of CRA International, Inc., is available at .

Page 1

November 19, 2014

American Financial Services Association

Table of Contents

1. BACKGROUND........................................................................................................................... 8

2. APPROACH AND METHODOLOGY....................................................................................... 8 2.1. DATA ................................................................................................................................. 9

3. THE RETAIL AUTOMOTIVE FINANCE MARKET ................................................................ 9 3.1. MARKET SIZE ................................................................................................................. 10 3.2. MARKET SEGMENTS ...................................................................................................... 11

3.2.1. Direct and Indirect Channels...................................................................................................... 11 3.2.2. New and Used Vehicles.............................................................................................................. 12 3.2.3. Prime, Non-Prime or Subprime Consumers ............................................................................ 13 3.2.4. Lease or Purchase ...................................................................................................................... 14

3.3. MARKET PARTICIPANTS ................................................................................................. 15

3.3.1. Consumers ................................................................................................................................... 15 3.3.2. Types of Financial Institutions.................................................................................................... 17

3.4. MARKET CONCENTRATION............................................................................................. 17 3.5. SEGMENT SPECIALIZATION ............................................................................................ 20 3.6. SOURCES OF CAPITAL.................................................................................................... 20 3.7. COMPETITIVE FACTORS ................................................................................................. 22

3.7.1. Online Credit Application Networks ? the efficient auction .................................................... 24 3.7.2. Dealers .......................................................................................................................................... 25 3.7.3. Vehicle Manufacturers ................................................................................................................ 31 3.7.4. Bringing it all together ? transactions........................................................................................ 32

4. FAIR LENDING COMPLIANCE FOR INDIRECT AUTOMOTIVE FINANCE .................. 34

Page 2

November 19, 2014

American Financial Services Association

4.1. BACKGROUND ................................................................................................................ 34 4.2. ECOA ? DISPARATE IMPACT AND TREATMENT.............................................................. 38 4.3. FAIR LENDING AND DEALER RESERVE ........................................................................... 41 4.4. IDENTIFYING RACE AND ETHNICITY FOR VEHICLE PURCHASES................................... 45

4.4.1. Vehicle purchases by Race/Ethnicity Shares .......................................................................... 48 4.4.2. Specific Proxy Methods .............................................................................................................. 51 4.4.3. Testing of Specific Proxy Methods ............................................................................................ 54

4.5. MEASUREMENT OF DEALER RESERVE ......................................................................... 63 4.6. COMPLEXITY OF THE TRANSACTION.............................................................................. 64

4.6.1. Unknown consumer-specific factors ......................................................................................... 65 4.6.2. Unknown dealer-specific factors................................................................................................ 67

4.7. PRICING DIFFERENCES ACROSS DEALERS ................................................................... 70 4.8. OBSERVABILITY OF DEALER CONTRACTS .................................................................... 71

5. OBSERVED PRICES IN THE CURRENT MARKET........................................................... 72 5.1. OBSERVED CONTRACT RATES AND BUY RATES.......................................................... 73 5.2. SIMILARLY SITUATED CONSUMERS AND CONTROLS ................................................... 76

5.2.1. Adjusting for proxy bias............................................................................................................... 77 5.2.2. Deal Specific Controls................................................................................................................. 79 5.2.3. Unknown Factors ......................................................................................................................... 80

6. ALTERNATIVE DEALER COMPENSATION MODELS ..................................................... 81 6.1. CFPB AND DOJ PREFERENCES ................................................................................... 82 6.2. TESTING.......................................................................................................................... 82

6.2.1. Scenario 1..................................................................................................................................... 83 6.2.2. Scenario 2..................................................................................................................................... 83

Page 3

November 19, 2014

American Financial Services Association

6.2.3. Scenario 3..................................................................................................................................... 84 6.2.4. Scenario 4..................................................................................................................................... 84

7. CONCLUSIONS AND RECOMMENDATIONS.................................................................... 85 8. APPENDIX A. PROJECT TEAM ............................................................................................ 88 9. APPENDIX B. GLOSSARY OF TERMS AND ACRONYMS............................................. 90 10. APPENDIX C. HOUSEHOLD VEHICLE OWNERSHIP BY STATE AND

RACE/ETHNICITY, 2010-2012............................................................................................... 94 11. APPENDIX D. BISG ASSUMPTIONS USED BY CHARLES RIVER ASSOCIATES FOR

THIS STUDY ........................................................................................................................... 100 12. APPENDIX E. RACE/ETHNICITY PROXIES: DIFFERENCES BETWEEN BISG

CALCULATIONS: CRA V. CFPB ........................................................................................ 101 13. APPENDIX F. BISG 2-WAY TABLES, "HEAT-MAPS" ..................................................... 105 14. APPENDIX G. BISG FALSE POSITIVES AND NEGATIVES BY TRACT, FICO,

INCOME, AND LMI ................................................................................................................. 107 15. APPENDIX H. CRA CONTRACT DATA VARIABLES ...................................................... 111 16. APPENDIX I. CRA CONTRACT DATA DESCRIPTIVE STATISTICS ........................... 112 17. APPENDIX J. DEALER RESERVE REGRESSION RESULTS ...................................... 128 18. APPENDIX K. COST/BENEFIT SCENARIO RESULTS................................................... 132 19. APPENDIX L. CHARTS AND TABLES ............................................................................... 140 20. APPENDIX M. REFERENCES ............................................................................................. 141

Page 4

November 19, 2014

American Financial Services Association

INTRODUCTION

Over the past few years, regulatory focus on fair lending examination of the indirect automotive finance market has increased significantly. Recent regulatory developments that impact the indirect auto finance market include the issuance on March 21, 2013 of the Consumer Financial Protection Bureau (CFPB) Bulletin 201302,1 "Indirect Auto Lending and Compliance with the Equal Credit Opportunity Act" (Bulletin) which details the manner in which certain policies related to dealer discretion have the potential to create significant fair lending risks for financial institutions that participate in this important consumer market.2 At the same time, methodologies used by regulatory agencies for fair lending examinations have changed significantly. For example, the CFPB issued "Using Publicly Available Information to Proxy for Unidentified Race and Ethnicity" (White Paper) on September 17, 2014 which presents its methodology for using a proxy to assign race/ethnicity to consumers obtaining auto financing.

In this research, we illustrate the complexities of indirect automobile financing and evaluate current regulatory fair lending practices observed in the industry. The research uses data collected from a number of market participants and aggregated in order to inform the discussions concerning dealer compensation, prices observed in the market, and the costs and benefits to consumers of alternative dealer compensation methods.

Highlights of the study include demonstrating that:

? The markets for purchasing automobiles (the retail automotive market) and for financing automobiles (the automotive finance market) are complex, highly interconnected and highly competitive.

? Accurately analyzing dealer reserve is difficult for a number of reasons, and failure to consider these challenges increases the potential for drawing erroneous conclusions.

? The methods commonly used by regulators to proxy race and ethnicity, including the recently applied Bayesian Improved Surname Geocoding

1 CFPB Bulletin 2013-02, March 21, 2013. 2 In this paper we use the term `financial institution' to refer to any company that finances new or used vehicle sales. Financial institutions include banks, non-banks, credit unions, captive and non-captive companies, direct lenders and indirect finance companies, and buyhere pay-here dealers.

Page 5

November 19, 2014

American Financial Services Association

(BISG) method, are conceptually flawed in their application and subject to significant bias and estimation error.

? The use of biased race and ethnicity proxies creates significant measurement errors, which likely result in overstated disparities and overstatements of alleged consumer harm.

? The Department of Justice (DOJ) recognizes that dealer reserves depend on objective, observable business factors. Failure to consider legitimate business factors for observed disparities increases the potential for reaching erroneous conclusions.

? Aggregating contracts originated by individual dealers to the portfolio level may create the appearance of differential pricing on a prohibited basis when none exists.

? When appropriately considering the relevant market complexities and adjusting for proxy bias and error, the observed variations in dealer reserve are largely explained.

? Alternative dealer compensation structures, such as "flats," may lead to increased borrowing costs for many minority and non-minority consumers and, in turn, may limit access to credit for some consumers.

A first step in designing an appropriate fair lending strategy is developing the conceptual framework. The intricacies of this very complex market require more complex strategies than those used to date. Given the realities of the regulatory landscape and the limited tools available for analysis, the ability to perform meaningful, accurate and actionable analyses of dealer reserves at the portfolio level is very circumscribed. Based on our analysis, we offer the following key recommendations:

? In calculating any disparities at the portfolio level, make adjustments to the population to: o Exclude any volumes from dealers with zero dealer reserve. o Exclude any volumes from dealers with no variance in reserve. o Exclude any dealers with counts insufficient to monitor dealer activity ? specifically, exclude dealers with fewer than 2 contracts from a protected class member and 2 contracts from non-Hispanic whites and a total of 5 contracts. (Similar restrictions should be applied when analyzing for age or gender).

? Implement economic controls to adjust for general economic conditions beyond the control of the financial institution or dealer. Specifically, adjust for:

Page 6

November 19, 2014

American Financial Services Association

o Location ? the analyses should include MSA level fixed effect controls. Market demand/supply conditions vary by MSA.

o New/Used ? these markets are completely different on many dimensions and the negotiation around trade in values may directly impact dealer reserves.

o Broad credit tranche ? this is not equivalent to controlling for credit score in the buy rate analysis but rather recognizes that prime and subprime markets vary broadly.

o Month of origination. ? Adjust for the known bias in the use of the BISG proxy methodology:

o If using a continuous approach, determine the "count" of affected minority consumers by applying a threshold after the application of the continuous method. At the very least, the consumers with BISG probabilities less than 50% should not be included in any calculation of consumer harm.

o Require verification/certification that any consumer receiving settlement funds or other remediated responses actually is a member of a protected class.

o If funds remain in the settlement fund, these should revert to the financial institution and not become part of any regulatory "settlement fund."

? When applying the BISG method, use a stricter threshold for any actions taken prior to 2012. The BISG approach had never been used historically, no one would have used it for monitoring, and applying a recent innovation to past behavior is unfair to financial institutions. For all originations prior to 2011, a 70% BISG threshold, or similar, should be applied.

? Going forward, while financial institutions may, given sufficient volumes, monitor activity quarterly, no remediation should take place until the end of the year. This will help adjust for seasonality during an annual cycle.

? The analysis should include a dealer level focus. There must be adjustments for the aggregation issue.

? The continuous BISG methodology should not be used in any analysis of indirect auto underwriting. The econometric interpretation of such a result is overly difficult.

Page 7

November 19, 2014

American Financial Services Association

1. BACKGROUND

Historically, most research on fair lending has followed the focus of regulatory enforcement on discrimination in mortgage markets, and far less research or supervisory activity involved the automotive retail market. To assist in filling the research void that exists, this study provides examination of the following:

? the size and scope of the vehicle finance market. ? the history and evolution of indirect auto finance. ? the Equal Credit Opportunity Act (ECOA) and disparate treatment and impact. ? the history, applicability and accuracy of proxy analysis, including BISG. ? quantitative analysis of current pricing practices in the vehicle finance market. ? the identification and quantitative analysis of factors potentially impacting dealer

participation. ? the identification and quantitative analysis of alternative dealer compensation

methods, including an assessment of whether such alternative dealer compensation methods are likely to adversely impact the availability of credit for protected classes and lower income groups.

2. APPROACH AND METHODOLOGY

The research focuses on answering the following key questions:

1. What is the automotive finance market and how does it function?

2. Are there fair lending concerns with dealer discretion and dealer reserve?

a. Can these concerns be reliably addressed?

i. What are the challenges and how may they be addressed?

ii. What information is needed for financial institutions and dealers to monitor fair lending risk?

iii. What dealer reserve prices are observed in the market and what explains variations in those prices?

b. What are the advantages and disadvantages of particular methodologies?

3. What are alternative dealer compensation structures and how would they impact consumers' cost and access to credit, as well as other market participants including dealers and financial institutions?

Page 8

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download